Integrate PUC Lua's utf8 library into Luanti

CMakeLists.txt

@@ -290,6 +290,7 @@ if(NOT USE_LUAJIT)
 	add_subdirectory(lib/bitop)
 endif()
 add_subdirectory(lib/sha256)
+add_subdirectory(lib/lutf8)
 
 if(BUILD_UNITTESTS OR BUILD_BENCHMARKS)
 	add_subdirectory(lib/catch2)

doc/lua_api.md

@@ -11972,6 +11972,23 @@ Functions: bit.tobit, bit.tohex, bit.bnot, bit.band, bit.bor, bit.bxor, bit.lshi
 
 See http://bitop.luajit.org/ for advanced information.
 
+UTF-8 Library
+-------------
+
+Constants:
+
+* `utf8.charpattern`
+
+Functions:
+
+* `utf8.char(...)`
+* `utf8.codes(s [, lax])`
+* `utf8.codepoint(s [, i [, j [, lax]]])`
+* `utf8.len(s [, i [, j [, lax]]])`
+* `utf8.offset(s, n [, i])`
+
+See [the Lua 5.4 reference manual](https://www.lua.org/manual/5.4/manual.html#6.5) for more information.
+
 Tracy Profiler
 --------------
 

games/devtest/mods/unittests/misc.lua

@@ -341,3 +341,10 @@ local function test_ipc_poll(cb)
 	print("delta: " .. (core.get_us_time() - t0) .. "us")
 end
 unittests.register("test_ipc_poll", test_ipc_poll)
+
+local function test_utf8()
+	assert(#string.char(0xc3, 0xa4):match(utf8.charpattern) == 2)
+	assert(("\0"):match(utf8.charpattern) == "\0")
+	assert(utf8.char(0x11, 0x22, 0x10abcd) == string.char(0x11, 0x22, 0xf4, 0x8a, 0xaf, 0x8d))
+end
+unittests.register("test_utf8", test_utf8)

lib/lutf8/CMakeLists.txt

@@ -0,0 +1,5 @@
+add_library(lutf8 STATIC lutf8.c)
+target_include_directories(lutf8 PRIVATE ${LUA_INCLUDE_DIR})
+
+set(LUA_UTF8_INCLUDE_DIR ${CMAKE_CURRENT_SOURCE_DIR} PARENT_SCOPE)
+set(LUA_UTF8_LIBRARY lutf8 PARENT_SCOPE)

lib/lutf8/lutf8.c

@@ -1,25 +1,21 @@
-/*
-** $Id: lutf8lib.c $
-** Standard library for UTF-8 manipulation
-** See Copyright Notice in lua.h
-*/
+// PUC Lua UTF-8 library, with minor modifications for integration in Luanti.
+// Taken from https://github.com/lua/lua/blob/c15543b9afa31ab5dc564511ae11acd808405e8f/lutf8lib.c
+// MIT-licensed, see LICENSE.txt
 
 #define lutf8lib_c
 #define LUA_LIB
 
-#include "lprefix.h"
-
+#include "lutf8.h"
 
 #include <assert.h>
 #include <limits.h>
 #include <stdlib.h>
 #include <string.h>
+#include <stdint.h>
 
 #include "lua.h"
 
 #include "lauxlib.h"
-#include "lualib.h"
-#include "llimits.h"
 
 
 #define MAXUNICODE	0x10FFFFu
@@ -34,6 +30,9 @@
 #define iscontp(p)	iscont(*(p))
 
 
+typedef uint32_t l_uint32;
+typedef uint64_t lua_Unsigned;
+
 /* from strlib */
 /* translate a relative string position: negative means back from end */
 static lua_Integer u_posrelat (lua_Integer pos, size_t len) {
@@ -99,11 +98,11 @@ static int utflen (lua_State *L) {
   while (posi <= posj) {
     const char *s1 = utf8_decode(s + posi, NULL, !lax);
     if (s1 == NULL) {  /* conversion error? */
-      luaL_pushfail(L);  /* return fail ... */
+      lua_pushnil(L);  /* return fail ... */
       lua_pushinteger(L, posi + 1);  /* ... and current position */
       return 2;
     }
-    posi = ct_diff2S(s1 - s);
+    posi = (size_t)(s1 - s);
     n++;
   }
   lua_pushinteger(L, n);
@@ -137,17 +136,37 @@ static int codepoint (lua_State *L) {
     s = utf8_decode(s, &code, !lax);
     if (s == NULL)
       return luaL_error(L, MSGInvalid);
-    lua_pushinteger(L, l_castU2S(code));
+    lua_pushinteger(L, code);
     n++;
   }
   return n;
 }
 
+#define UTF8BUFFSZ 8
+
+// Taken from https://github.com/lua/lua/blob/c15543b9afa31ab5dc564511ae11acd808405e8f/lobject.c#L385-L400
+static int luaO_utf8esc(char *buff, l_uint32 x) {
+  int n = 1;  /* number of bytes put in buffer (backwards) */
+  if (x < 0x80)  /* ascii? */
+    buff[UTF8BUFFSZ - 1] = (char)(x);
+  else {  /* need continuation bytes */
+    unsigned int mfb = 0x3f;  /* maximum that fits in first byte */
+    do {  /* add continuation bytes */
+      buff[UTF8BUFFSZ - (n++)] = (char)(0x80 | (x & 0x3f));
+      x >>= 6;  /* remove added bits */
+      mfb >>= 1;  /* now there is one less bit available in first byte */
+    } while (x > mfb);  /* still needs continuation byte? */
+    buff[UTF8BUFFSZ - n] = (char)((~mfb << 1) | x);  /* add first byte */
+  }
+  return n;
+}
 
 static void pushutfchar (lua_State *L, int arg) {
   lua_Unsigned code = (lua_Unsigned)luaL_checkinteger(L, arg);
   luaL_argcheck(L, code <= MAXUTF, arg, "value out of range");
-  lua_pushfstring(L, "%U", (long)code);
+  char bf[UTF8BUFFSZ];
+  int len = luaO_utf8esc(bf, (l_uint32)code);
+  lua_pushlstring(L, &bf[UTF8BUFFSZ - len], len);
 }
 
 
@@ -180,7 +199,7 @@ static int byteoffset (lua_State *L) {
   size_t len;
   const char *s = luaL_checklstring(L, 1, &len);
   lua_Integer n  = luaL_checkinteger(L, 2);
-  lua_Integer posi = (n >= 0) ? 1 : cast_st2S(len) + 1;
+  lua_Integer posi = (n >= 0) ? 1 : (lua_Integer)(len + 1);
   posi = u_posrelat(luaL_optinteger(L, 3, posi), len);
   luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 3,
                    "position out of bounds");
@@ -210,7 +229,7 @@ static int byteoffset (lua_State *L) {
     }
   }
   if (n != 0) {  /* did not find given character? */
-    luaL_pushfail(L);
+    lua_pushnil(L);
     return 1;
   }
   lua_pushinteger(L, posi + 1);  /* initial position */
@@ -239,8 +258,8 @@ static int iter_aux (lua_State *L, int strict) {
     const char *next = utf8_decode(s + n, &code, strict);
     if (next == NULL || iscontp(next))
       return luaL_error(L, MSGInvalid);
-    lua_pushinteger(L, l_castU2S(n + 1));
-    lua_pushinteger(L, l_castU2S(code));
+    lua_pushinteger(L, (lua_Integer)(n + 1));
+    lua_pushinteger(L, (lua_Integer)code);
     return 2;
   }
 }
@@ -267,7 +286,7 @@ static int iter_codes (lua_State *L) {
 
 
 /* pattern to match a single UTF-8 character */
-#define UTF8PATT	"[\0-\x7F\xC2-\xFD][\x80-\xBF]*"
+#define UTF8PATT	"[%z-\x7F\xC2-\xFD][\x80-\xBF]*"
 
 
 static const luaL_Reg funcs[] = {
@@ -276,14 +295,12 @@ static const luaL_Reg funcs[] = {
   {"char", utfchar},
   {"len", utflen},
   {"codes", iter_codes},
-  /* placeholders */
-  {"charpattern", NULL},
   {NULL, NULL}
 };
 
 
-LUAMOD_API int luaopen_utf8 (lua_State *L) {
-  luaL_newlib(L, funcs);
+LUALIB_API int luaopen_utf8 (lua_State *L) {
+  luaL_register(L, LUA_UTF8LIBNAME, funcs);
   lua_pushlstring(L, UTF8PATT, sizeof(UTF8PATT)/sizeof(char) - 1);
   lua_setfield(L, -2, "charpattern");
   return 1;

lib/lutf8/lutf8.h

@@ -0,0 +1,6 @@
+#pragma once
+
+#include "lua.h"
+
+#define LUA_UTF8LIBNAME "utf8"
+LUALIB_API int luaopen_utf8(lua_State *L);

src/CMakeLists.txt

@@ -573,6 +573,7 @@ include_directories(SYSTEM
 	${GMP_INCLUDE_DIR}
 	${JSON_INCLUDE_DIR}
 	${LUA_BIT_INCLUDE_DIR}
+	${LUA_UTF8_INCLUDE_DIR}
 	# on Android, Luanti depends on SDL2 directly
 	# on other platforms, only IrrlichtMt depends on SDL2
 	"$<$<PLATFORM_ID:Android>:${SDL2_INCLUDE_DIRS}>"
@@ -695,6 +696,7 @@ if(BUILD_CLIENT)
 		${GMP_LIBRARY}
 		${JSON_LIBRARY}
 		${LUA_BIT_LIBRARY}
+		${LUA_UTF8_LIBRARY}
 		sha256
 		${FREETYPE_LIBRARY}
 		${PLATFORM_LIBS}
@@ -785,6 +787,7 @@ if(BUILD_SERVER)
 		${JSON_LIBRARY}
 		${LUA_LIBRARY}
 		${LUA_BIT_LIBRARY}
+		${LUA_UTF8_LIBRARY}
 		sha256
 		${GMP_LIBRARY}
 		${PLATFORM_LIBS}

src/script/cpp_api/s_base.cpp

@@ -29,6 +29,7 @@ extern "C" {
 #else
 	#include "bit.h"
 #endif
+#include "lutf8.h"
 }
 
 #include <cstdio>
@@ -84,6 +85,10 @@ ScriptApiBase::ScriptApiBase(ScriptingType type):
 	lua_pushstring(m_luastack, LUA_BITLIBNAME);
 	lua_call(m_luastack, 1, 0);
 
+	// Load utf8 library
+	lua_pushcfunction(m_luastack, luaopen_utf8);
+	lua_call(m_luastack, 0, 0);
+
 #if BUILD_WITH_TRACY
 	// Load tracy lua bindings
 	tracy::LuaRegister(m_luastack);

src/script/cpp_api/s_security.cpp

@@ -3,6 +3,7 @@
 // Copyright (C) 2013 celeron55, Perttu Ahola <celeron55@gmail.com>
 
 #include "cpp_api/s_security.h"
+#include "log.h"
 #include "lua_api/l_base.h"
 #include "filesys.h"
 #include "porting.h"
@@ -38,6 +39,7 @@ static void shallow_copy_table(lua_State *L, int from=-2, int to=-1)
 	if (from < 0) from = lua_gettop(L) + from + 1;
 	if (to   < 0) to   = lua_gettop(L) + to   + 1;
 	lua_pushnil(L);
+	assert(lua_istable(L, from));
 	while (lua_next(L, from) != 0) {
 		assert(lua_type(L, -1) != LUA_TTABLE);
 		// duplicate key and value for lua_rawset
@@ -96,6 +98,7 @@ void ScriptApiSecurity::initializeSecurity()
 		"table",
 		"math",
 		"bit",
+		"utf8",
 		// Not sure if completely safe. But if someone enables tracy, they'll
 		// know what they do.
 #if BUILD_WITH_TRACY
@@ -296,6 +299,7 @@ void ScriptApiSecurity::initializeSecurityClient()
 		"table",
 		"math",
 		"bit",
+		"utf8",
 		// Not sure if completely safe. But if someone enables tracy, they'll
 		// know what they do.
 #if BUILD_WITH_TRACY